LIVE · TUE, JUL 14, 2026 --:--:-- ET
Issue Nº 84 COST TOTAL $14828.20 ARTICLES TODAY 14 TOKENS TOTAL 9.52B
Podcast
aiexpert
Subscribe · $8/mo
Running the wire
Market Open-weight models now 10x cheaper than frontier APIs, closing capability gap to 7 months Policy Q-Day countdown: NIST finalizes three post-quantum crypto standards; organizations must test by 2030 Funding Together AI closes $800M Series C at $8.3B; Aramco Ventures leads as open-model infra consolidates Policy Google, Microsoft, Salesforce, NVIDIA launch ARD spec for agent tool discovery and governance Chips NVIDIA Blackwell GB300 NVL72 hits $0.123/million tokens, 35x cheaper than Hopper, 50x per-watt gain Market Chamath: AI token costs double every 45 days; productivity gains plateau at 5% max Chips China chip exports surge 96% YoY to $177B in H1 2026, driven by memory price boom Policy Demis Hassabis calls for U.S.-led AI standards body to test frontier models Breaking Databricks showcases Lakebase as high-velocity stateful memory for agentic AI workflows Research Prime Intellect releases Verifiers v1; DAG trace format eliminates quadratic data growth in agentic RL Funding TerraFirma closes $115M for remote-operated construction; SpaceX, Anduril among backers Chips Intel invests €5B in Ireland fab to boost Xeon 6, Intel 3 production Policy New York enacts one-year data center moratorium for projects over 50 megawatts; first US state ban Funding Cyera raises $600M at $12B valuation on AI agent security infrastructure play Funding Cyera Raises $600M at $12B Valuation, Quadrupling Worth in 18 Months on AI Security Demand Chips NVIDIA halves authorized customer list in Asia to combat chip smuggling into China Chips AMD Zen 6 debut July 22–23 with EPYC Venice; 2nm, 256-core, 70% perf boost Breaking OpenAI Codex reaches 7 million active users; temporarily removes 5-hour usage cap Funding Secondary markets open late-stage private equity to retail: WHOOP secondary case study Chips Researchers develop programmable thermal material that remembers state without power Market Open-weight models now 10x cheaper than frontier APIs, closing capability gap to 7 months Policy Q-Day countdown: NIST finalizes three post-quantum crypto standards; organizations must test by 2030 Funding Together AI closes $800M Series C at $8.3B; Aramco Ventures leads as open-model infra consolidates Policy Google, Microsoft, Salesforce, NVIDIA launch ARD spec for agent tool discovery and governance Chips NVIDIA Blackwell GB300 NVL72 hits $0.123/million tokens, 35x cheaper than Hopper, 50x per-watt gain Market Chamath: AI token costs double every 45 days; productivity gains plateau at 5% max Chips China chip exports surge 96% YoY to $177B in H1 2026, driven by memory price boom Policy Demis Hassabis calls for U.S.-led AI standards body to test frontier models Breaking Databricks showcases Lakebase as high-velocity stateful memory for agentic AI workflows Research Prime Intellect releases Verifiers v1; DAG trace format eliminates quadratic data growth in agentic RL Funding TerraFirma closes $115M for remote-operated construction; SpaceX, Anduril among backers Chips Intel invests €5B in Ireland fab to boost Xeon 6, Intel 3 production Policy New York enacts one-year data center moratorium for projects over 50 megawatts; first US state ban Funding Cyera raises $600M at $12B valuation on AI agent security infrastructure play Funding Cyera Raises $600M at $12B Valuation, Quadrupling Worth in 18 Months on AI Security Demand Chips NVIDIA halves authorized customer list in Asia to combat chip smuggling into China Chips AMD Zen 6 debut July 22–23 with EPYC Venice; 2nm, 256-core, 70% perf boost Breaking OpenAI Codex reaches 7 million active users; temporarily removes 5-hour usage cap Funding Secondary markets open late-stage private equity to retail: WHOOP secondary case study Chips Researchers develop programmable thermal material that remembers state without power
Chips

NVIDIA Blackwell GB300 NVL72 hits $0.123/million tokens, 35x cheaper than Hopper, 50x per-watt gain

NVIDIA published latest inference cost benchmarks for its Blackwell Ultra GB300 NVL72 platform on the NVIDIA developer blog and at GTC 2026, claiming $0.123 per million tokens at 116 tokens-per-second per user interactivity using TensorRT-LLM and Dynamo software. This 35x cost reduction versus Hopper (H100) sets a new floor for low-latency production agentic AI workloads and establishes per-token cost—not raw FLOPS or power draw alone—as the only metric that matters for inference TCO.

The GB300 NVL72 achieves this through full-stack codesign: 72 Blackwell Ultra GPUs, 288GB HBM3e per GPU, and a unified 130 TB/s NVLink fabric that acts as a single rack-scale system. For mixture-of-experts models like DeepSeek-R1 and DeepSeek V4 Pro, the platform delivers 50x higher throughput per megawatt versus Hopper. Continuous software optimization is a key multiplier: software tweaks alone have yielded 4x performance improvements on GB200 over three months, with further gains as TensorRT-LLM kernels are optimized for open-source and closed-model inference.

NVIDIA frames the metric as 'revenue per megawatt': a $5M investment in a GB200 NVL72 system can generate $75M in token revenue, implying a 15x ROI. This economics-first positioning is aimed at inference service providers (CoreWeave, DeepInfra, Together AI, Fireworks AI) who operationalize token costs immediately. For a power-constrained world, the implication is that hardware generations now improve on a single variable: how many high-value tokens per watt.

For infrastructure teams, this codifies what many suspected: the Blackwell generation ended the GPU-scarcity era and began the token-cost optimization race. Inference providers who have not yet optimized software stacks (DynoSim, quantization strategies, expert routing) on Blackwell are leaving real margin on the table. The threshold for entry-level agentic AI inference is now the ability to run production-grade MoE models at <$0.15/MT margins.

Sources

Get the signal before the noise.